PROJECT ABSTRACT The clinical diagnosis of Alzheimer's disease (AD) is based on the core criterion of memory impairment, but biomarker-detectable pathophysiological changes start decades before clinical symptoms. This preclinical phase, in which someone has the neuropathology of AD but does not yet show clinical symptoms, is crucial for potential intervention and timely diagnosis for patient and caregiver. The preclinical phase is currently only detectable with expensive or invasive biomarkers. Because current cognitive measures are not sensitive to the preclinical phase of AD and have low specificity and large variation with regard to individuals' educational and cultural exposure, there is a critical need to develop sensitive, low-cost, and high-access cognitive markers for early detection in diverse older adults. The primary goal of this project is to investigate if novel psycholinguistic metrics of existing cognitive test data can accurately identify people in the earliest stages of AD. The semantic fluency task?naming as many animals in one minute?tests semantic memory, one of the first cognitive domains to become impaired in AD. Traditionally, semantic fluency is scored by the total number of items. However, there is a wealth of information at the item-level of this task, because words are organized in a semantic network that becomes vulnerable during AD, specifically for words that are poorly connected and not often used. These traits of words in the semantic network can be captured with novel psycholinguistic metrics, such as lexical frequency, e.g., `dog' is a high-frequent word in our language, as opposed to `iguana.' Nine novel item-level psycholinguistic metrics of semantic fluency have been selected to: investigate how AD imaging biomarkers in nondemented adults relate to psycholinguistic measures (Aim 1, K99); estimate the temporality of semantic impairment across the AD continuum (Aim 2, R00); and determine the sensitivity and specificity of psycholinguistic measures to predict progression to clinical AD (Aim 3, R00). Since the relationship of demographics to psycholinguistic metrics is not well understood, this project strives to deconstruct cultural and demographic effects on these metrics in order to maximize their potential utility in early diagnosis among diverse older adults. The project will employ advanced statistical analyses to investigate data from three large longitudinal cohorts with diverse participants and semantic fluency data in English, Spanish, and Dutch. This K99/R00 proposal lays the foundation for an independent research program focused on semantic processing in normal aging and across the AD continuum. The proposed project will provide the applicant with 1) new training in computational semantics and structural equation modeling, 2) experience with imaging biomarker data, and 3) a strong foundation in cultural neuropsychology. These experiences will supplement the applicant's strong background in Neurolinguistics and Epidemiology. The results of the proposed research have the potential to translate into a clinical tool that we can use across educationally, linguistically, and culturally diverse individuals to refine who to select for intervention trials.